Before I get into how the models did this weekend, I thought I’d put in my quick two cents on the replacement referees. While everyone is complaining about the calls, a couple of people have pointed out that things aren’t as out of whack as they might seem. After week 2, for example, ESPN stats noted that the number and types of calls are basically the same as they were last year. Brian Burke also noted that, despite the perception that the refs were being biased towards the home teams, home teams are not winning by unusual amounts so far. These are important things to keep in mind, because it provides some evidence that the replacement refs aren’t completely crazy.
On the other hand, Brian is right to say “This is not to say that the replacement refs are not influenced by home crowds or that they’re doing a good or bad job”. Because it isn’t so much the direct outcomes of games or the number and type of the calls that people are complaining about, it’s the quality of the calls. It doesn’t matter if there are usually 4 pass interference calls (or whatever the number is) in a game and the replacement refs call 4 a game; it matters which four plays they call them on. If they miss four real penalties and throw the flag on four plays where nothing happened, the replacement refs are doing a bad job. That’s something we can only tell by watching the games and getting the opinions of people who know what penalties should be, not by looking at aggregate stats.
And, as far as the Seattle-Green Bay ending last night; I don’t know for sure that ‘real’ refs would have gotten that call right, but that sure was a doozy. They missed offensive pass interference and it looked to me like the defender had possession before Tate grabbed the ball on the way down. But if the officials decided it was a simultaneous catch, that goes to the offense and it’s a touchdown. As any fan of the UFC knows, “don’t leave it up to the officials”. The defender should have just knocked it down, or if he saw the Titans-Lions game, he could have knocked it way back out of the end zone. The refs blew that call, but the Packers didn’t put the game away either. The Seahawks get some credit for keeping it close and they got some help to pull out the win.
On to the prediction results. All three models were 8-8 picking the over/under, ignoring any potential ‘close calls’ like Luigi’s prediction of 47.5 for the Titans-Lions game against a number of 47. At some point I’ll actually see if these things are worthwhile at any level, like if they predict well if the prediction is far away from the number. But for now, Luigi is 13-18-1, Yoshi 1 is 16-15-1, and Yoshi 2 is 13-18-1.
Speaking of the Lions, I’m on board with TMQ (and Brian again) on this one. They should have gone for it, just with a good play. When the Lions lined up to go for it, I was surprised but happy. In a game where you might not get the ball back (it would be over if Tennessee scored, or if they just managed to run out the remaining six minutes or so), and particularly when the Titans have scored in pretty much every way imaginable, and even more when your starting QB is out, I think you need to go for the win. I was less happy when the Lions ran a QB sneak against a stacked front that was waiting for exactly that kind of thing. It was apparently a mistake, but I wonder what they would have called if they actually tried to run a play. I would vote for some manner of screen.
In terms of outright winners, Luigi went 7-9 while the Yoshis went 8-8. That puts Luigi at 12-20, Yoshi 1 at 17-15, and Yoshi 2 at 15-17. Luigi usually does better than this, so it’s a tiny bit concerning (he at least shouldn’t be below chance). On the other hand, if you used the SBR line as your way to pick who should win, you would only be 16-16 at this point. It seems like maybe there have been more upsets than you would think so far. So maybe things will turn around as we move forward. At any rate, as is always more interesting we can look at the moneyline picks. Luigi had a pick for every game except for the Ravens-Pats and went 5-10. That doesn’t sound good, but his five hits were underdogs which means he would have come out slightly ahead if he bet $100 on each game, primarily thanks to Minnesota and Kansas City. Yoshi 1 picked every game except for Houston-Denver and did even better, going 9-6 on all underdogs. Yoshi 2 passed on Giants-Panthers and Indy-Jacksonville and went 7-7 also all on underdogs. So all of the models had decent turnouts, but the Yoshis really did well. Through two weeks of picks Luigi is 8-22 and solidly in the hole, Yoshi 1 is 15-16 and well ahead, and Yoshi 2 is 11-18 and a bit ahead.
The Vikings were probably the big surprise winner of the weekend, handling the 49ers. I was surprised, but each of the models thought they had a good chance and were potentially even the slightest of favorites. Why might that be? It’s kind of backwards for a team with Adrian Peterson, but the Vikes throw the ball well and aren’t so hot at running. And, despite the raves for Alex Smith so far, the 49ers still aren’t an amazing passing team (although they are good at running). So overall, the two were relatively well-matched on offense; you can see some of the numbers in a nice table at Brian’s site. From the same page, you can see that the teams are also pretty well-matched on defense; if anything, the Vikes are a tiny bit better. Also, the Vikings were at home. And thus an ‘upset’ is born.
Alright, on to the spreads. Looking at every game, Luigi went 6-9-1 (the Jets’ overtime field goal made that game a push). Yoshi 1 had a career week at 12-3-1, and Yoshi 2 was 7-8-1. Across the two weeks that puts the models at 12-17-3, 20-19-3, and 13-16-3 respectively. More reasonably, the models would avoid some games if they could. In those games the models were 5-7-1, 9-3-1, and 5-7-1 respectively. So only Yoshi 1 came out ahead this week. Across two weeks Luigi is 10-15-3 in ‘pickable’ games, Yoshi 1 is 17-8-2, and Yoshi 2 is 10-14-3. Again it looks like Yoshi 1 is leading the pack, but if the games so far have been turning out a little oddly then it might not stay this way. I would be a little sad if it turned out you should just expect the home team to win by 3 all the time.
I think I’ll debut my power rankings after next week, just to add one more game to help even things out. But for the curious, my top 5 are Houston, Baltimore, the Giants, Atlanta, and New England. My bottom 5 are Tennessee, St. Louis, Jacksonville, Cleveland, and in a stunner, the Packers. Green Bay has not been playing well so far. Just ahead of the Pack? The Eagles.
On to Bill Simmons. He counts every game, so I do too. He has slightly different lines, so the results might be a bit different from what I have above; when I compared Luigi to the Bovada lines in the past, Bill’s lines turned out much easier to beat. This week, Luigi went 6-10, Yoshi 1 10-6, Yoshi 2 6-10, and Bill went 9-7. Across two weeks Luigi is at 12-18-2, Yoshi 1 18-12-2, Yoshi 2 12-18-2, and Bill 17-13-2. Bill’s doing pretty well so far, but picks are tight. He had the Jets at -2.5, so the field goal was a winner for him while all my models had Miami and lost. Sometimes it’s just that half a point.
And as usual I’ll end with the Hilton SuperContest. Luigi got Tampa and Minnesota and a push on Miami for a 2-2-1 week and an overall record of 4-4-2 and 5 points. Yoshi 1 got Tampa, Arizona, KC, and a push with Miami for a 3-1-1 week (missing on St. Louis) and an overall record of 6-3-1 for 6.5 points. Yoshi 2 got Minnesota, Tampa, Arizona, and a push on Miami (missing Washington) for a 3-1-1 week also and an overall record of 5-3-2 and 6 points. If you add in the 3 points I got for my week 1 picks, the models are in a range from 8 to 9.5 points. That’s where the bulk of the crowd is right now while the leader is at 13.
To tantalize you a bit for next weekend: I think my most unlikely pick is either Buffalo over New England or Jets over 49ers. More football on Thursday!